deploying Galaxy for the first time
Dear small scale Galaxy Admins, I am planning to deploy a production Galaxy Server to be used by the researchers at my research institute. I have been following the tutorials at Galaxy Training! <https://training.galaxyproject.org/training-material/learning-pathways/admin-training.html> and some questions have come to mind: * *OS version*: Ubuntu 22.04 vs 24.04?: since the tutorial is a bit old already (based on Ubuntu 18.04), I assume I should stick to the latest stable Ubuntu version. * Should I install the *latest stable Galaxy* version (25.0 at this moment)? * Should Galaxy server be *on a VM or on a physical node*? My HPC infrastructure consists of login and management nodes working as VMs inside a Proxmox 7.X server, and a bunch of computing nodes (there's a couple dedicated storage nodes). If Galaxy and the PostgreSQL database are stored on the same VM and it has enough resources it would be quite easy to manage, since you can reboot it very quickly and it's possible and easy to take snapshots of the VM. * If the Galaxy Server is stored on a VM on the Proxmox host, I would not be able to access the VM directly from outside. Which changes should I make to access the Galaxy? Is there any alternative to redirect traffic to port 80 or 8080 from the Proxmox host to the VM? I’ve experienced some issues with Nginx if I do this, since Gunicorn uses a gunicorn.sock file instead. If I add iptables rules so that the proxmox server redirects all the traffic from its port 80 to the port 80 of the VM, I get the following error at nginx: |nginx: 2025/06/13 10:55:55 [crit] 46578#46578: *1 connect() to unix:/srv/galaxy/var/config/gunicorn.sock failed (2: No such file or directory) while connecting to upstream, client: 10.X.X.X, server: 192.168.2.X, request: "GET / HTTP/1.1", upstream: "http://unix:/srv/galaxy/var/config/gunicorn.sock:/", host: "10.X.X.X", referrer: "http://10.X.X.X/"| I've read I should use NGINX on the Proxmox host if I don't want to do port-forwarding with iptables. Alternatively the VPN from the research institute could be reconfigured or redirect directly to the VM. Do you have any thoughts about this? If this is not the right channel to ask, please let me know. Thanks a lot in advance, Cheers, Juan
Hi Juan These are excellent questions! I hope you can join us on Thursday for our meeting (https://galaxyproject.org/events/2025-06-small-scale/) which will be the best place to get a broad range of answers and tips for your questions. In the meantime, I suggest you add your points to the topic suggestions for Thursday (https://docs.google.com/document/d/1KTz4qxZBG4Z3sF23dVKPWXYz1mCNo2YDygSUHqtA...) If you can't attend this week, we will start answering your questions on the mailing list and/or suggest to put them onto the admin channel on element. Regards, Hans-Rudolf Hans-Rudolf On 16.06.2025 21:41, Juan Antonio Cordero Varela wrote:
Dear small scale Galaxy Admins,
I am planning to deploy a production Galaxy Server to be used by the researchers at my research institute.
I have been following the tutorials at Galaxy Training! <https:// training.galaxyproject.org/training-material/learning-pathways/admin- training.html> and some questions have come to mind:
*
*OS version*: Ubuntu 22.04 vs 24.04?: since the tutorial is a bit old already (based on Ubuntu 18.04), I assume I should stick to the latest stable Ubuntu version.
*
Should I install the *latest stable Galaxy* version (25.0 at this moment)?
*
Should Galaxy server be *on a VM or on a physical node*? My HPC infrastructure consists of login and management nodes working as VMs inside a Proxmox 7.X server, and a bunch of computing nodes (there's a couple dedicated storage nodes). If Galaxy and the PostgreSQL database are stored on the same VM and it has enough resources it would be quite easy to manage, since you can reboot it very quickly and it's possible and easy to take snapshots of the VM.
*
If the Galaxy Server is stored on a VM on the Proxmox host, I would not be able to access the VM directly from outside. Which changes should I make to access the Galaxy? Is there any alternative to redirect traffic to port 80 or 8080 from the Proxmox host to the VM? I’ve experienced some issues with Nginx if I do this, since Gunicorn uses a gunicorn.sock file instead. If I add iptables rules so that the proxmox server redirects all the traffic from its port 80 to the port 80 of the VM, I get the following error at nginx: |nginx: 2025/06/13 10:55:55 [crit] 46578#46578: *1 connect() to unix:/srv/ galaxy/var/config/gunicorn.sock failed (2: No such file or directory) while connecting to upstream, client: 10.X.X.X, server: 192.168.2.X, request: "GET / HTTP/1.1", upstream: "http://unix:/srv/ galaxy/var/config/gunicorn.sock:/", host: "10.X.X.X", referrer: "http://10.X.X.X/"|
I've read I should use NGINX on the Proxmox host if I don't want to do port-forwarding with iptables. Alternatively the VPN from the research institute could be reconfigured or redirect directly to the VM. Do you have any thoughts about this?
If this is not the right channel to ask, please let me know.
Thanks a lot in advance,
Cheers,
Juan
_______________________________________________ Small-scale-admins mailing list -- small-scale-admins@lists.galaxyproject.org To unsubscribe send an email to small-scale-admins-leave@lists.galaxyproject.org
participants (2)
-
Hans-Rudolf Hotz -
Juan Antonio Cordero Varela